The Cognitive Distortion Journal (CDJ) is a smart journaling tool that helps remedy distorted thinking. It can feel impossible to follow the CBT technique of labeling distorted thinking and finding alternative modes of thought (i.e. reframing) while cognitive distortions are occurring. The CDJ does that work for you. -- The CDJ is in beta testing!!
Config currently assumes user knowledge of OpenAI API and available GPT models. After talking to several people about AI/LLM it appears most common knowledge (generally outside of tech) does not go past ChatGPT.
This issue can be used to suggest GPT models. This can be done in a few ways, but first the models I recommend from my usage of the app are gpt-4-1106-preview for analyses and gpt-4 for chat. While 1106 gpt-4 is slower than 1106 gpt-3.5, it produces higher quality analyses. For chat, gpt-4 is far better and producing conversational messages than any other I've tried so far which produces too much content in general. This makes it incredibly fast and it gets to the point. I've noticed it's more socratic in its therapy approach, which is more helpful.
One of the ways this can be done is by auto-populating the config with the suggestions if the field is empty. This might seem a little too pushy though for people who know these models. Instead a better approach would be to ornament the fields or create a Suggest button that they may press to auto-populate the fields instead.
It may be useful to provide a guide as well on how to retrieve an API key from their OpenAI account. It should also be mentioned how their API key safely encrypted before it is stored in the database.
Config currently assumes user knowledge of OpenAI API and available GPT models. After talking to several people about AI/LLM it appears most common knowledge (generally outside of tech) does not go past ChatGPT.
This issue can be used to suggest GPT models. This can be done in a few ways, but first the models I recommend from my usage of the app are
gpt-4-1106-previewfor analyses andgpt-4for chat. While 1106 gpt-4 is slower than 1106 gpt-3.5, it produces higher quality analyses. For chat, gpt-4 is far better and producing conversational messages than any other I've tried so far which produces too much content in general. This makes it incredibly fast and it gets to the point. I've noticed it's more socratic in its therapy approach, which is more helpful.One of the ways this can be done is by auto-populating the config with the suggestions if the field is empty. This might seem a little too pushy though for people who know these models. Instead a better approach would be to ornament the fields or create a Suggest button that they may press to auto-populate the fields instead.
It may be useful to provide a guide as well on how to retrieve an API key from their OpenAI account. It should also be mentioned how their API key safely encrypted before it is stored in the database.